Improved AIS data simplification algorithm for extracting typical routes considering motion continuity

Area of Science:

• Maritime technology
• Data science
• Artificial intelligence

Background:

• Raw Automatic Identification System (AIS) data is redundant, hindering maritime applications like navigation and route prediction.
• Overlapping and repetitive vessel routes in AIS data present challenges for direct analysis.
• Extracting typical vessel trajectories from complex AIS datasets requires effective simplification methods.

Purpose of the Study:

• To propose an improved algorithm for simplifying Automatic Identification System (AIS) data.
• To extract typical vessel routes while maintaining movement continuity.
• To evaluate the impact of data simplification on deep learning models for maritime applications.

Main Methods:

• Developed an enhanced distance threshold pruning technique.
• Analyzed vessel operation continuity to address route segment discontinuities.
• Applied the simplification algorithm to Automatic Identification System (AIS) data.

Main Results:

• Simplified AIS data significantly improved training efficiency and prediction accuracy in deep learning trajectory forecasting.
• Enhanced anomaly detection capabilities with fewer false positives.
• Models trained on simplified data showed faster convergence and higher precision compared to those trained on raw data.

Conclusions:

• The proposed simplification method offers practical advantages for real-time maritime monitoring.
• The algorithm is a valuable tool for improving operational efficiency in maritime applications.
• Simplified AIS data enhances the performance of deep learning models for trajectory prediction and anomaly detection.